Brand Lift 360
Book a 15-Minute Call

Resources · Methodology

The AI Visibility Methodology We Ran on Ourselves First

By Danzell Nunsuch, Co-founder, Chief Strategy & Innovation Officer ·

Brand Lift 360 measures AI visibility in five separate layers: entity accuracy, answer authority, citation authority, recommendation presence and demand capture. Being cited by AI is not the same as being recommended, and a recommendation is not a booked appointment, so each layer is scored on its own evidence. We ran the method on our own firm first, and the results are below.

Get Your Free AI Visibility Audit

Free · Five prioritized improvements within two business days

Doctor at her clinic office desk reviewing results on a tablet
Five layers, scored separatelyEight audit phasesTested on Brand Lift 360 first

At a glance

The Short Version

Four things to know before you read a single score.

  • AI visibility is not one number. We measure five layers separately and report each on its own evidence.
  • A citation means an answer used a page. A mention means it named the business. A recommendation means it presented the business as a provider worth considering. A booking is a separate customer action. None of them proves the next one happened.
  • Every finding carries a dated evidence record — the raw answer, the source, the observation, and its limits.
  • We ran the methodology on Brand Lift 360 before offering it to anyone: 63 baseline observations, named in 2, cited in 1. This page shows that worked example in full.

Four different events

The Distinction That Guides the Work

Most AI visibility reporting collapses four different events into one score. We record them independently:

  1. 01

    Citation

    The AI answer used the page as a source

  2. 02

    Mention

    The answer named the business

  3. 03

    Recommendation

    The answer presented the business as a provider worth considering

  4. 04

    Booking

    The customer acted — a separate, business-side outcome

A practice can be cited without being named, named without being recommended, and recommended without a single additional booking if its phones and forms cannot catch the demand. Each gap has a different fix, which is why each is measured on its own layer.

What we measure

The Five Layers

Each layer checks one thing and has one primary measure. The first is entity accuracy: an entity is your practice as search engines and AI systems identify it, one business with one set of facts attached to it.

  1. Layer 1: Entity Accuracy

    What it checks
    Correct facts, profiles, and technical access
    Primary measure
    Verified accuracy score — 10 checks
    How it is scored

  2. Layer 2: Answer Authority

    What it checks
    Substantive answers and original evidence on your own site
    Primary measure
    Weighted coverage (70) and quality (30) score
    How it is scored

  3. Layer 3: Citation Authority

    What it checks
    Accurate corroboration on external pages
    Primary measure
    Fixed-panel corroboration score
    How it is scored

  4. Layer 4: Recommendation Presence

    What it checks
    Whether answers present you as a provider worth considering
    Primary measure
    Recommendation rate within the measured sample
    How it is scored

  5. Layer 5: Demand Capture

    What it checks
    Whether inquiries become booked appointments
    Primary measure
    Readiness score, reported separately from business outcomes
    How it is scored

Layer by layer

How Each Layer Is Scored

Layer 1

Entity Accuracy

Verified accuracy score — 10 checks

Ten equally weighted checks: business identity, location and service area, contact routes, hours, category and services, people and credentials, Google Business Profile, external profiles, technical accessibility, and structured information. A wrong phone number or a blocking access issue scores zero on its check and gets immediate triage. A check that was not assessed is reported as not assessed — never as zero, and never silently passed.

Layer 2

Answer Authority

Weighted coverage (70) and quality (30) score

We freeze a universe of priority buying questions for the practice — the questions patients actually ask, checked against real demand evidence, not invented. Coverage (70 points) scores whether a substantive answer exists for each question; quality (30 points) scores checkable expertise, readable and technically accessible text, and original, useful information. One strong page may answer several questions; the method does not reward multiplying thin pages for minor wording differences.

Layer 3

Citation Authority

Fixed-panel corroboration score

From the baseline answers, we extract the external pages AI actually used, select up to ten of the most-cited relevant sources, and freeze that panel for the 90-day cycle. Each panel page is scored for accurate corroboration. The panel never changes mid-cycle to make a score look better: a page that disappears keeps its weight and is recorded as absent. Your own site's citations are reported separately — they belong to Layer 2's story, not this one.

Layer 4

Recommendation Presence

Recommendation rate within the measured sample

The percentage of valid answers to the approved commercial question set that present the practice as a provider worth considering, classified as Recommended, Mentioned, Discouraged, or Absent — for the practice and the same three named competitors, on the same questions. The standard wave runs 30 approved prompts across four declared surfaces (a surface is one place AI answers appear, such as an AI assistant or an AI search feature), repeated on three separate days. We report the numerator, the denominator, and the range across repeats. It describes the measured sample — not market share, and not every answer AI gives.

Layer 5

Demand Capture

Readiness score, reported separately from business outcomes

Two outputs, always kept separate: a readiness score (do calls route and get answered, do forms deliver and create a record, does booking work on mobile, is first response owned and measured) and observed business outcomes. Improving a form does not prove AI created demand, and this layer never claims it does.

The audit, step by step

How an Audit Runs: Eight Phases

Each phase has one defined output. The free AI Visibility Audit is the baseline layer of this method.

Get your free AI Visibility Audit
  1. 01

    Market definition

    Signed scope: one entity, one location, services, geography, three named competitors

  2. 02

    Prompt universe

    Approved questions with intent, cluster, and stable IDs

  3. 03

    Baseline collection

    Saved raw answers, collection settings, execution status

  4. 04

    Entity audit

    Authoritative facts, source checks, technical access evidence

  5. 05

    Answer audit

    Question-to-page mapping, quality checks, retrieval evidence

  6. 06

    Citation analysis

    Frozen baseline panel; emerging sources tracked separately

  7. 07

    Recommendation benchmark

    Results for the practice and the same competitors, same questions

  8. 08

    Demand capture audit

    Tested inquiry paths and reconciled business outcomes

The baseline is the humbling part, and it is supposed to be. When we ran this method on Brand Lift 360, the answers named us rarely and cited our own site once — while every one of our visibility wins came from our own pages and no third-party source named us at all. That is an entity and corroboration problem, not a content-volume problem, and it set our entire fix order: directories and profiles first, page openings second, reviews third, content fourth. A methodology that cannot embarrass its own firm is not measuring anything.

From our experience · Danzell Nunsuch

How findings are recorded

The Evidence Standard

Every material finding carries an evidence ID, an observed date, a source URL or retained artifact, the exact observation, a reviewer, and its limitation.

Raw answers and dated page captures are stored with the engagement. Missing data is reported as unavailable. Nothing is inferred from an inaccessible page or a missing export.

Every evidence record

  1. 01Evidence ID
  2. 02Observed date
  3. 03Source URL or retained artifact
  4. 04Exact observation
  5. 05Reviewer
  6. 06Limitation

Observations, interpretations, and proposed actions are reported in separate categories:

  • Observation“the phone number differs on two profiles”
  • Hypothesis“this may contribute to inconsistent entity resolution”
  • Intervention“correct the profiles and remeasure”

Worked example

Worked Example: Our Own Baseline

Wave 1, Brand Lift 360, measured with the same method we run for practices:

  • 63observations across the declared AI surfaces
  • 2Named in 2 of the 63
  • 1Cited in 1 of the 63

A second, directional snapshot

3 of 20answers named Brand Lift 360

A second, directional snapshot on October 26, 2026 (10 buying questions × 2 AI surfaces, 20 answers) named Brand Lift 360 in 3 of 20 — every one sourced from brandlift360.com itself. No third-party page named us in any answer. The most-named firm in that snapshot was named 5 times.

Re-measurement, in public

  1. Wave 2
  2. Wave 3
  3. Wave 4

Re-measurement waves are scheduled for November 26, 2026, December 28, 2026, and January 26, 2027, on the same question set and settings, and the results will be published on this page whether they flatter us or not.

Limits

What This Method Does Not Prove

Read every score on this page with these four limits in mind.

  1. A citation is an observed input to an answer, not a complete view of how an AI engine retrieves information.

  2. Recommendation Presence describes a curated, monitored sample of questions. Repeated sampling shows variability; it does not make the sample representative of all patients.

  3. Scores are house standards, calibrated through pilots. They are decision tools, not certifications, and no score on this page is a promise of future rankings, citations, or bookings.

  4. Competitor tracking provides context. It is not a control group, and we do not report it as one.

Questions

Frequently Asked Questions

How the method compares, what the free audit covers and what it cannot promise.

Get your free AI Visibility Audit
How is this different from an SEO audit?

An SEO audit measures how pages rank in search results. This methodology measures what AI answers say — whether they have your facts right, whether they use your pages, whether outside sources corroborate you, whether they recommend you by name, and whether the resulting inquiries get booked. The two overlap at the entity and content layers and diverge from there.

What does the free AI Visibility Audit include?

The free audit runs the buying questions in your specialty across the declared AI surfaces and reports where your practice is named, cited, or absent, with the five prioritized improvements the full methodology would address first. It is the baseline layer of this method, offered free.

How long does a full measurement cycle run?

Ninety days: a baseline wave, then remeasurement at Day 30, Day 60, and Day 90 on the frozen question set and citation panel, so every change is measured against the same standard.

Why publish your own low scores?

Because a measurement firm's first obligation is to show its instrument works on the firm itself. Named in 2 of 63 is our baseline, published with the method that produced it and the dates we will re-measure. If the numbers move, the same page will show it.

Do you guarantee a practice will be recommended by AI?

No. The methodology measures and prioritizes; it does not guarantee rankings, citations, recommendations, or bookings, and any firm that guarantees AI placement is selling something it cannot control.

References

Sources

  1. Brand Lift 360 AI Visibility Methodology v1.4 (frozen September 29, 2026) — internal standard, summarized here in full for the measurement definitions.
  2. Wave 1 baseline workbook, Brand Lift 360 (2026).
  3. AI visibility baseline snapshot, October 26, 2026.

Keep Reading

AEO GlossaryAI Search Visibility ServiceHow AI Assistants Recommend PracticesCase Studies

Your next step

See Where Your Practice
Stands in AI Search

Start with a free AI Visibility Audit: the baseline layer of this method, run on your practice.

Get Your Free AI Visibility AuditBook a 15-Minute Call

Prefer to call? (916) 597-0086

Free audit · No obligation

What the free audit covers

  1. 01

    Your buying questions

    The questions patients in your specialty ask, run across the declared AI surfaces.

  2. 02

    Named, cited or absent

    Where your practice appears in the answers, and where it does not.

  3. 03

    Five priorities

    The five prioritized improvements the full methodology would address first.